<h1 id="introduction-rk6p2w8fn">Introduction</h1>This is a V-PRED version of the <a rel="ugc" href="https://civarchive.com/models/1240179/yesmix-xl-illustrious">YesmixXL model</a>. It was trained on the same dataset as YesmixXL but uses v_prediction sampling instead of the standard SDXL epsilon method. The model was fine-tuned on <a rel="ugc" href="https://civarchive.com/models/833294/noobai-xl-nai-xl">NoobAI-V-PRED</a> using v_prediction method and further DPO-optimized with a <a rel="ugc" href="https://huggingface.co/datasets/zakp/DPO_MERGE_DATASET">personal anime preference dataset</a> for enhanced output quality. Keys for <code>v_prediction</code> and <code>zero_terminal_snr</code> are integrated, allowing direct image generation in ComfyUI. A known minor issue is a rare blue artifact in some image regions, resolvable by regenerating; otherwise, the model demonstrates stable generation across diverse prompts.<h1 id="usage-yshrzuvxe">Usage</h1>Sampler: Euler A (Recommended)CFG Scale: 6~7.5 (Recommended)Steps: 25 (Recommended)Single Prompt Weight: &lt;1.3, 1 is recommended

Files with SHA256: a2d3521d9cd5299404166b6fe097c64e98838a3592b45aed234dc6bd7f5b3057